Synchronizing Photomultiplier ADC Output With 3-Axis Controller Position Data in 3D Scanner

(warning: software engineer)
I'm working pro-bono on an experimental setup redesign.  This is (essentially) a 3d scanner.  We are collecting signal from a photomultiplier using an off the shelf ADC.  At the same time we are driving a sample on a 3-axis mount underneath the scan head.  The whole setup us controlled by a general purpose x86 system (my domain).Unfortunately, we've hit a performance ceiling due to the complexity of synchronizing the controller position with the ADC signal.  In other words, right now we are guaranteeing adc/position synchronization by starting/stopping the adc at each point.  I've hit the EBSCO stacks because I'm sure this problem has been solved before but it's either too trivial to show up in print or too complex to have a published solution (probably the first).TLDR; I'm looking for a way to synchronize adc output with positional data from the controller.  I can get strong position/time points using the computers HPET but I'm at a loss for getting time-encoded ADC data.  
I'm open (excited) about the possibility of internalizing the ADC, if we can find the right unit.
Thank you for your time,-A

alexIt would seem to me that you may benefit from time-quantising your input signal In other words taking periodic samples that allow you to process the data and once completed, only then look for the next data. In truth this is exactly what the ADC does. You just have to synchornise the result to your output.  This does mean that a smooth outline that you are scanning will become linearised between readings, so of course you want as short a conversion time as possible. You can certainly improve the sampling rate if you push the data into a FIFO (or even LIFO) buffer, but if intention is for the 3d head to move in synchronicity with the scan then there are 2 limiting factors- the speed of response of the 3d head compared to the photomultiplier scan, and of course the conversion time of the ADC.First I would define the fastest rate that you can work at (in truth I would start at a much slower rate to prove the system). I would use a sample/hold amplifier on the output of the photomultipler. It will sample according to the rate defined above. Once the sample/hold is in the hold state, the ADC is initiated and on completion of the conversion your calculations are processed and the signal is output to the 3D head. Now the process can repeat.If you can buffer the data, then the slower (I presume) 3D head can trundle along working on the stored data, but I would still reccomend the S/H.

What about having the ADC sample at regular intervals, and periodically steer the photomultiplier to a known reference position (full dark or full light), allowing the samples to be registered to the scan position after the fact? 

Extend the ADC's output word length by additional bits, and stick position or time information in the additional bits. Read the extended word into the computer, and use software to separate the pieces. Use gray-coded time or position, so as to get correct reading if the time or position changes at the moment it is sampled.